arctic Posted April 12, 2006 Report Share Posted April 12, 2006 http://getswiftfox.com/ Try it. I am posting from it already. Not bad. A special package for athlons, semprons, durons and intel processors. Now check this memory usage: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7025 fabian 15 0 88352 29m 16m S 2.0 3.2 0:04.60 firefox-bin This is with the Mandriva default Firefox 1.07 package: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7264 fabian 15 0 97.9m 30m 18m S 1.0 3.3 0:03.59 mozilla-firefox Quote Link to comment Share on other sites More sharing options...
ffi Posted April 12, 2006 Report Share Posted April 12, 2006 I don´t know about these builds but on Windows the optimised moox builds actually performed worse than the standard build: http://www.howtocreate.co.uk/browserSpeed.html Quote Link to comment Share on other sites More sharing options...
daniewicz Posted April 12, 2006 Report Share Posted April 12, 2006 (edited) Artic: I use the KDE System Guard to see memory usage, and have never seen the "m" you show. Is "m" for million? Could I write the swiftfox memory usage as 88352 = 0.088352m? If I am interpreting this "m" correctly, then the swiftfox usage is extraordinarily small compared to the mandriva rpm. Yet the other numbers RES and SHR are similar. What are those? I tried swiftfox a year ago or so and found that it was slower than the version downloaded directly from mozilla.org. I used the scragz test (attached) to measure speed. I also tried to build my own firefox, but could never match the speed of the mozilla.org version. edit: I was not allowed to upload the scragz test because it is an xhtml file. Edited April 12, 2006 by daniewicz Quote Link to comment Share on other sites More sharing options...
iphitus Posted April 12, 2006 Report Share Posted April 12, 2006 (edited) %MEM is what matters... and that's 0.1% different... which could be due to a range of variables. There's too many variables to consider what you post reliable or even somewhat useful results. Anyway, that package looks like extra rice to me. vrmmm vrmmm omgoptomised. There's bigger bottlenecks than for what architecture the binary is compiled for. Not to mention the lack of benchmarks, or at least if they are there, well hidden benchmarks. Just adds to that rice factor. Edited April 12, 2006 by iphitus Quote Link to comment Share on other sites More sharing options...
arctic Posted April 12, 2006 Author Report Share Posted April 12, 2006 I don't see it as a "ricer" thing but as a nice side-project for those users who do not boast 1GB Ram on their machines. Computers with only e.g. 128 MB Ram will surely benefit from such a project. I admit, it is not much of reduced memory, but the project at least shows that it is possible to reduce the amount of Ram needed by Firefox by optimized coding. Not everyone who has a low-Ram machine will want to use Dillo or Links as webbrowsers. :) PS: The 3.2 / 3.3 % Mem in my statistics does not look like much as I ran "top" on a 1GB Ram system. Make the same test on a 128 MB or 256 MB Ram machine and it might make a real difference. Quote Link to comment Share on other sites More sharing options...
daniewicz Posted April 12, 2006 Report Share Posted April 12, 2006 Well I like the idea of compiling firefox with optimizations for particular CPU's. I have just never had any success producing a speed increase as measured using synthetic benchmarks. Quote Link to comment Share on other sites More sharing options...
scarecrow Posted April 12, 2006 Report Share Posted April 12, 2006 Well I like the idea of compiling firefox with optimizations for particular CPU's. I have just never had any success producing a speed increase as measured using synthetic benchmarks. Neither did I- and I had compiled quite a few Firefox betas. Maybe there WAS a difference... but it wasn't noticeable. Quote Link to comment Share on other sites More sharing options...
tyme Posted April 12, 2006 Report Share Posted April 12, 2006 i've been using the firefox 2.0 beta for a few days now and it seems to run better than 1.5, but i haven't bothered with swifterfox. Quote Link to comment Share on other sites More sharing options...
iphitus Posted April 13, 2006 Report Share Posted April 13, 2006 I don't see it as a "ricer" thing but as a nice side-project for those users who do not boast 1GB Ram on their machines. Computers with only e.g. 128 MB Ram will surely benefit from such a project. I admit, it is not much of reduced memory, but the project at least shows that it is possible to reduce the amount of Ram needed by Firefox by optimized coding. Not everyone who has a low-Ram machine will want to use Dillo or Links as webbrowsers. :) PS: The 3.2 / 3.3 % Mem in my statistics does not look like much as I ran "top" on a 1GB Ram system. Make the same test on a 128 MB or 256 MB Ram machine and it might make a real difference. the 0.1% is such a minor difference, that it can easily be attributed to errors and uncontrolled variables, which was pretty much my point above :) James Quote Link to comment Share on other sites More sharing options...
Steve Scrimpshire Posted April 13, 2006 Report Share Posted April 13, 2006 (edited) A .1% decrease in memory and a 1.0% increase in cpu usage? Sounds pointless to me. Here's mine: swiftfox PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4376 omar 15 0 2276 1076 892 S 0.0 0.2 0:00.00 firefox 4379 omar 19 0 2824 1368 888 S 0.0 0.3 0:00.00 run-mozilla.sh 4384 omar 15 0 91256 33m 16m S 0.0 7.1 0:04.26 firefox-bin ------------------------------------------------------------------------- firefox PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4269 omar 15 0 2788 1348 904 S 0.0 0.3 0:00.00 mozilla-firefox 4274 omar 23 0 2816 1364 888 S 0.0 0.3 0:00.00 run-mozilla.sh 4279 omar 15 0 89080 38m 16m S 0.0 8.2 0:04.42 mozilla-firefox ----------------------------------------------------------------------------- swiftfox PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4492 omar 18 0 2272 1076 892 S 0.0 0.2 0:00.00 firefox 4495 omar 18 0 2820 1368 888 S 0.0 0.3 0:00.00 run-mozilla.sh 4500 omar 16 0 83008 33m 16m S 0.0 7.2 0:03.50 firefox-bin -------------------------------------------------------------------------------- firefox PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4582 omar 15 0 2788 1348 904 S 0.0 0.3 0:00.00 mozilla-firefox 4587 omar 24 0 2820 1364 888 S 0.0 0.3 0:00.00 run-mozilla.sh 4592 omar 16 0 112m 39m 16m S 0.0 8.4 0:06.47 mozilla-firefox 1.2% on my 512MB system. 6.144 MB....not worth the hassle in my opinion. Edited April 13, 2006 by Steve Scrimpshire Quote Link to comment Share on other sites More sharing options...
arctic Posted April 13, 2006 Author Report Share Posted April 13, 2006 Okay... I shut up. :D Quote Link to comment Share on other sites More sharing options...
daniewicz Posted April 13, 2006 Report Share Posted April 13, 2006 I think swiftfox is being compiled to optimize speed, not minimize memory usage or CPU load. If we are going to quantify swiftfox benefits (or lack thereof), we should be looking at rendering speed. Quote Link to comment Share on other sites More sharing options...
iphitus Posted April 13, 2006 Report Share Posted April 13, 2006 And the time it takes from url entry to displaying a page, is mostly bottlenecked by the network, not the rendering. Better changes can be made elsewhere. Quote Link to comment Share on other sites More sharing options...
daniewicz Posted April 13, 2006 Report Share Posted April 13, 2006 Well I certainly won't argue with that iphitus. You are correct. Not directly related to your comment, but I have been able to increase the speed of web browsing by simply using a hosts file to sidestep things like ad.doubleclick.net (and its multitude of brethren). It is amazing the number of servers that are contacted with a simple click on your favorite web site. Using ntop has opened my eyes to this and my hosts file is now one of my closest friends. B) Quote Link to comment Share on other sites More sharing options...
iphitus Posted April 14, 2006 Report Share Posted April 14, 2006 (edited) Ditto. I use an ad blocking hosts file and the fasterfox extension to speed up my browser. ad blocking hosts file stops those ads downloading, so that saves a lot of bw/time. One extension worth giving a shot is fasterfox. It does things to improve those bandwidth bottlenecks such as link prefetching, heavier caching, more connections, no rendering delay, and other things. Unlike other improvements, this one has technical merit, and is logical in how it works. Makes a damned nice difference. http://fasterfox.mozdev.org James. Edited April 14, 2006 by iphitus Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.